icu_provider_source 2.2.0

# This file is part of ICU4X. For terms of use, please see the file
# called LICENSE at the top level of the ICU4X source tree
# (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ).

# These sentence boundary rules are based on UAX #29, Unicode Version 17.0.0.
# https://www.unicode.org/reports/tr29/tr29-47.html

segmenter_type = "sentence"

[[tables]]
name = "CR"

[[tables]]
name = "LF"

[[tables]]
name = "Extend"

[[tables]]
name = "Sep"

[[tables]]
name = "Format"

[[tables]]
name = "Sp"

[[tables]]
name = "Lower"

[[tables]]
name = "Upper"

[[tables]]
name = "OLetter"

[[tables]]
name = "Numeric"

[[tables]]
name = "ATerm"

[[tables]]
name = "SContinue"

[[tables]]
name = "STerm"

[[tables]]
name = "Close"

[[tables]]
# SB5
name = "Lower"
left = "Lower"
right = "Extend"

[[tables]]
# SB5
name = "Lower"
left = "Lower"
right = "Format"

[[tables]]
# SB5
name = "Upper"
left = "Upper"
right = "Extend"

[[tables]]
# SB5
name = "Upper"
left = "Upper"
right = "Format"

[[tables]]
# SB5
name = "OLetter"
left = "OLetter"
right = "Extend"

[[tables]]
# SB5
name = "OLetter"
left = "OLetter"
right = "Format"

[[tables]]
# SB5
name = "ATerm"
left = "ATerm"
right = "Extend"

[[tables]]
# SB5
name = "ATerm"
left = "ATerm"
right = "Format"

[[tables]]
# SB5
name = "STerm"
left = "STerm"
right = "Extend"

[[tables]]
# SB5
name = "STerm"
left = "STerm"
right = "Format"

[[tables]]
# SB5
name = "ATerm_Close"
left = "ATerm_Close"
right = "Extend"

[[tables]]
# SB5
name = "ATerm_Close"
left = "ATerm_Close"
right = "Format"

[[tables]]
# SB5
name = "ATerm_Close_Sp"
left = "ATerm_Close_Sp"
right = "Extend"

[[tables]]
# SB5
name = "ATerm_Close_Sp"
left = "ATerm_Close_Sp"
right = "Format"

[[tables]]
# SB5
name = "STerm_Close"
left = "STerm_Close"
right = "Extend"

[[tables]]
# SB5
name = "STerm_Close"
left = "STerm_Close"
right = "Format"

[[tables]]
# SB5
name = "STerm_Close_Sp"
left = "STerm_Close_Sp"
right = "Extend"

[[tables]]
# SB5
name = "STerm_Close_Sp"
left = "STerm_Close_Sp"
right = "Format"

[[tables]]
# SB5
name = "Upper_ATerm"
left = "Upper_ATerm"
right = "Extend"

[[tables]]
# SB5
name = "Upper_ATerm"
left = "Upper_ATerm"
right = "Format"

[[tables]]
# SB5
name = "Lower_ATerm"
left = "Lower_ATerm"
right = "Extend"

[[tables]]
# SB5
name = "Lower_ATerm"
left = "Lower_ATerm"
right = "Format"

[[tables]]
# SB5
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Extend"

[[tables]]
# SB5
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Format"

[[tables]]
# SB7
name = "Upper_ATerm"
left = "Upper"
right = "ATerm"

[[tables]]
# SB7
name = "Lower_ATerm"
left = "Lower"
right = "ATerm"

[[tables]]
# SB8
name = "ATerm_Close"
left = "ATerm"
right = "Close"

[[tables]]
# SB8
name = "ATerm_Close"
left = "ATerm_Close"
right = "Close"

[[tables]]
# SB8
name = "ATerm_Close"
left = "Upper_ATerm"
right = "Close"

[[tables]]
# SB8
name = "ATerm_Close"
left = "Lower_ATerm"
right = "Close"

[[tables]]
# SB8
name = "ATerm_Close_Sp"
left = "ATerm"
right = "Sp"

[[tables]]
# SB8
name = "ATerm_Close_Sp"
left = "Upper_ATerm"
right = "Sp"

[[tables]]
# SB8
name = "ATerm_Close_Sp"
left = "Lower_ATerm"
right = "Sp"

[[tables]]
# SB8
name = "ATerm_Close_Sp"
left = "ATerm_Close"
right = "Sp"

[[tables]]
# SB8
name = "ATerm_Close_Sp"
left = "ATerm_Close_Sp"
right = "Sp"

[[tables]]
# SB8
name = "ATerm_Close_Sp_ParaSep"
left = "ATerm_Close"
right = "Sep"

[[tables]]
# SB8
name = "ATerm_Close_Sp_CR"
left = "ATerm_Close"
right = "CR"

[[tables]]
# SB8
name = "ATerm_Close_Sp_ParaSep"
left = "ATerm_Close"
right = "LF"

[[tables]]
# SB8
name = "ATerm_Close_Sp_ParaSep"
left = "ATerm_Close_Sp"
right = "Sep"

[[tables]]
# SB8
name = "ATerm_Close_Sp_CR"
left = "ATerm_Close_Sp"
right = "CR"

[[tables]]
# SB8
name = "ATerm_Close_Sp_ParaSep"
left = "ATerm_Close_Sp"
right = "LF"

[[tables]]
# SB8/SB9/SB10/SB11
name = "ATerm_Close_Sp_ParaSep"
left = "Lower_ATerm"
right = "Sep"

[[tables]]
# SB8/SB9/SB10/SB11
name = "ATerm_Close_Sp_CR"
left = "Lower_ATerm"
right = "CR"

[[tables]]
# SB8/SB9/SB10/SB11
name = "ATerm_Close_Sp_ParaSep"
left = "Lower_ATerm"
right = "LF"

[[tables]]
# SB8/SB9/SB10/SB11
name = "ATerm_Close_Sp_ParaSep"
left = "Upper_ATerm"
right = "Sep"

[[tables]]
# SB8/SB9/SB10/SB11
name = "ATerm_Close_Sp_CR"
left = "Upper_ATerm"
right = "CR"

[[tables]]
# SB8/SB9/SB10/SB11
name = "ATerm_Close_Sp_ParaSep"
left = "Upper_ATerm"
right = "LF"

[[tables]]
# SB8a/9
name = "STerm_Close"
left = "STerm"
right = "Close"

[[tables]]
# SB8a/9
name = "STerm_Close"
left = "STerm_Close"
right = "Close"

[[tables]]
# SB8a/9/10
name = "STerm_Close_Sp"
left = "STerm"
right = "Sp"

[[tables]]
# SB8a/9/10
name = "STerm_Close_Sp"
left = "STerm_Close"
right = "Sp"

[[tables]]
# SB8a/9/10
name = "STerm_Close_Sp"
left = "STerm_Close_Sp"
right = "Sp"

[[tables]]
# SB9/10
name = "STerm_Close_Sp_ParaSep"
left = "STerm_Close"
right = "Sep"

[[tables]]
# SB9/10
name = "STerm_Close_Sp_CR"
left = "STerm_Close"
right = "CR"

[[tables]]
# SB9/10
name = "STerm_Close_Sp_ParaSep"
left = "STerm_Close"
right = "LF"

[[tables]]
# SB9/10
name = "STerm_Close_Sp_ParaSep"
left = "STerm_Close_Sp"
right = "Sep"

[[tables]]
# SB9/10
name = "STerm_Close_Sp_CR"
left = "STerm_Close_Sp"
right = "CR"

[[tables]]
# SB9/10
name = "STerm_Close_Sp_ParaSep"
left = "STerm_Close_Sp"
right = "LF"

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm"
right = "Unknown"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "Upper_ATerm"
right = "Unknown"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "Lower_ATerm"
right = "Unknown"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close"
right = "Numeric"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close"
right = "Unknown"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp"
right = "Close"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp"
right = "Numeric"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp"
right = "Unknown"
interm_break_state = true

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Close"

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Format"

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Numeric"

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Sp"

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "Unknown"

[[tables]]
# SB8
name = "ATerm_Close_Sp_SB8"
left = "ATerm_Close_Sp_SB8"
right = "SContinue"

[[rules]]
# SB1
left = [ "sot" ]
right = [ "Any" ]
break_state = true

[[rules]]
# SB2
left = [
  "Sep",
  "Lower",
  "Upper",
  "OLetter",
  "Numeric",
  "Extend",
  "Format",
  "ATerm_Close",
  "ATerm_Close_Sp",
  "ATerm_Close_Sp_ParaSep",
  "ATerm_Close_Sp_CR",
  "STerm_Close",
  "STerm_Close_Sp",
  "STerm_Close_Sp_ParaSep",
  "STerm_Close_Sp_CR",
  "Lower_ATerm",
  "Upper_ATerm"
]
right = [ "eot" ]
break_state = true

[[rules]]
# SB3
left = [
  "CR",
  "ATerm_Close_Sp_CR",
  "STerm_Close_Sp_CR"
]
right = [ "LF" ]
break_state = false

[[rules]]
# SB4
left = [ "Sep", "CR", "LF" ]
right = [
  "ATerm",
  "Numeric",
  "Sep",
  "Lower",
  "Upper",
  "OLetter",
  "Sp",
  "Unknown",
  "STerm",
  "Close",
  "SContinue",
  "Format",
  "Extend",
  "CR"
]
break_state = true

[[rules]]
# SB4
left = [ "Sep", "LF", "ATerm_Close", "ATerm_Close_Sp" ]
right = [ "LF" ]
break_state = true

[[rules]]
# SB4
left = [
  "ATerm_Close_Sp_ParaSep",
  "ATerm_Close_Sp_CR",
  "STerm_Close_Sp_ParaSep",
  "STerm_Close_Sp_CR"
]
right = [
  "ATerm",
  "Numeric",
  "Lower",
  "Upper",
  "OLetter",
  "Sp",
  "Unknown",
  "STerm",
  "Close",
  "SContinue",
  "Format",
  "Extend"
]
break_state = true

[[rules]]
# SB6
left = [ "ATerm", "Upper_ATerm", "Lower_ATerm" ]
right = [ "Numeric" ]
break_state = false

[[rules]]
# SB7
left = [ "Upper_ATerm", "Lower_ATerm" ]
right = [ "Upper" ]
break_state = false

[[rules]]
# SB8
left = [ "ATerm", "ATerm_Close", "ATerm_Close_Sp" ]
right = [ "Lower" ]
break_state = false

[[rules]]
# SB8
left = [ "Lower_ATerm", "Upper_ATerm" ]
right = [ "Lower" ]
break_state = false

[[rules]]
# SB8
left = [ "ATerm_Close_Sp_SB8" ]
right = [ "Lower" ]
break_state = false

[[rules]]
# SB8a
left = [
  "ATerm",
  "ATerm_Close",
  "ATerm_Close_Sp",
  "STerm",
  "STerm_Close",
  "STerm_Close_Sp",
  "Lower_ATerm",
  "Upper_ATerm"
]
right = [ "SContinue", "ATerm", "STerm" ]
break_state = false

[[rules]]
# SB11
left = [
  "ATerm_Close_Sp_ParaSep",
  "ATerm_Close_Sp_CR",
  "STerm_Close_Sp_ParaSep",
  "STerm_Close_Sp_CR"
]
# NOTE(egg): left=Meow_CR, right=LF would be wrong, but it is prevented by another rule above.
right = [ "ATerm", "Lower", "OLetter", "Upper", "Numeric", "STerm", "CR", "LF", "Sep" ]
break_state = true

[[rules]]
# SB11
left = [ "ATerm_Close_Sp", "STerm_Close_Sp" ]
right = [ "Numeric", "Upper", "Lower", "Close", "Unknown", "OLetter" ]
break_state = true

[[rules]]
# SB11
left = [ "ATerm_Close", "STerm", "STerm_Close" ]
right = [ "Numeric", "Upper", "Lower", "OLetter", "Unknown" ]
break_state = true

[[rules]]
# SB11
left = [ "ATerm", "Upper_ATerm", "Lower_ATerm", "STerm" ]
right = [ "Unknown", "Upper", "Lower", "OLetter", "ATerm", "STerm" ]
break_state = true

[[rules]]
# SB998
left = [ "Any" ]
right = [ "Any" ]
break_state = false